Journal of Applied Ecology

2004

41

, 464–475

© 2004 British Ecological Society

Blackwell Publishing, Ltd.

Changes in species richness with stocking density of marine bivalves

H. A. BEADMAN, M. J. KAISER, M. GALANIDI, R. SHUCKSMITH and R. I. WILLOWS*

School of Ocean Sciences, University of Wales-Bangor, Menai Bridge, Gwynedd LL59 5AB, UK; and

*

Environment Agency, National Centre for Risk Analysis and Options Appraisal, Kings Meadow House, Kings Meadow Road, Reading RG1 8DQ, UK

Summary

1.

Monocultures of mussels might alter the infaunal benthic community of adjacentand interstitial sediments through provision of a complex habitat, input of organically richmaterial and larval removal through filter feeding. At a site of commercial seabed musselcultivation, we aimed to determine the effect of mussels on the infaunal community ofan intertidal mudflat at different spatial scales and under different stocking strategies.

2.

Mussels were laid at four different densities (2, 3, 5 and 7·5 kg m

−

2

) on 400-m

2

plotsin a 4

×

4 Latin square. Benthic samples were collected within and 10–100 m distantfrom the cultivation area

c

. 7 months prior to and 18 months after seeding the plots withblue mussels. Benthic community characteristics were related to initial seeding densityand to the actual surface area of mussels associated with each set of samples collectedwithin replicate plots.

3.

The presence of mussels significantly changed the occurrence of some species ofthe infaunal community within the cultivated area. The infaunal communities sup-ported fewer individuals and species than control treatments at all but the lowest musselcover.

4.

Species richness and the abundance of individuals per unit area also declined withincreased area of mussel cover. The abundance of cirratulids and amphipods declinedstrongly with increasing mussel surface area.

5.

Although the species composition and abundance of individual invertebrate specieswere altered by the presence of mussels, the distribution of individuals among speciesremained relatively unchanged.

6.

Synthesis and applications

. Overall, mussel beds changed the infaunal community,but the effects were localized (0–10 m) and not detectable at larger scales (10–100 m).Changes in benthic community composition could be reduced (but not eliminated) bylowering the stocking density of mussels to either 2 or 3 kg m

−

2

. Given the small edgeeffects associated with cultivated mussel beds, the use of larger mussel beds would bepreferable to many smaller mussel beds.

Key-words

: community change, ecological impact, large-scale experiment, musselcultivation,

Mytilus edulis

Journal of Applied Ecology

(2004)

41

, 464–475

Introduction

Intensive monoculture of marine biota occurs globallyand its contribution to marine-derived sources of pro-tein has increased steadily over the last three decades

(FAO 2003). However, marine aquaculture is associatedwith negative ecological and socio-economic impacts(Kaiser

et al

. 1998; Naylor

et al

. 2000). Given the socialand economic importance of the coastal zone (Costanza

et al

. 1997), there is a pressing need to quantify the eco-logical consequences of different management approachesto mariculture. To date, the impact of intensive cul-tivation systems on biodiversity has focused primarilyon terrestrial systems, often using appropriately designed

Correspondence: M. J. Kaiser, School of Ocean Sciences,University of Wales-Bangor, Menai Bridge, Gwynedd LL595AB, UK (e-mail michel.kaiser@bangor.ac.uk).

465

Species richness and bivalve stocking density

© 2004 British Ecological Society,

Journal of Applied Ecology

,

41

,464–475

experimental manipulations. Such agriculture systemsare associated with reductions in species richnessand, in some cases, fewer associated plant, arthropodand vertebrate assemblages (Chamberlain

et al

. 2000;Symstad, Siemann & Haarstad 2000; Marshall

et al

.2003). The ecological effects associated with the cultivationof fish or bivalves suspended from rafts have tended to bestudied at the scale of the cultivation unit or multiplesthereof, and have focused on issues relating primarilyto the carrying capacity of specific marine systems. Incontrast, the seabed cultivation of bivalve molluscs ismore analogous to the monoculture of terrestrial cropsin that the cultivated species replaces, to varying extents,the original biota and habitat. However, unlike terres-trial systems, monocultures of bivalves, such as mussels,oysters and clams, are rarely supplemented with artificialfertilizers or chemicals for pest control. Nevertheless,monocultures of bivalves often occur within or close toareas of nature conservation importance and have thepotential to change the composition of local inverte-brate assemblages and their dependent predators.

Mussels (Bivalvia: Mytilidae) are distributed glob-ally and are a conspicuous feature of intertidal habi-tats on both hard and soft substrata (Seed 1976). Theyform a key component of many marine communitiesand are often the dominant organisms in terms of theirbiomass (Seed 1976; Herman 1993). Mussels also createa secondary habitat, composed of layers of musselswith accumulated sediment, faeces, pseudofaeces andshell debris, that supports a highly diverse associatedcommunity (Tsuchiya & Nishihira 1985, 1986;Ragnarsson & Raffaelli 1999) that differs from that ofthe surrounding sediments (Commito 1987; Dittman1990; Guenther 1996; Ragnarsson & Raffaelli 1999).In terms of physical habitat structure, mussels provide acomplex habitat capable of harbouring a diverse assem-blage of associated flora and fauna (Seed & Suchanek1992). Biologically, the mussels provide an input of sed-iment and organic matter in the form of faeces andpseudofaeces (Tsuchiya 1980; Kautsky & Evans 1987)and remove fine particulate matter and some larvae ofbenthic invertebrates through filter-feeding activities(Cowden, Young & Chia 1984; Morgan 1992; Wahl 2001).Consequently, mussel communities have the capacityto either enhance or degrade the infaunal assemblage.

Mussels occur in naturally settled beds in the inter-tidal and subtidal zones. Alternatively, they can be laidin ‘artificial’ beds for cultivation. Mussels are cultivatedthroughout much of the world, including Europe,Asia and North America, with total mussel landings ofmore than 1·3 million tonnes in 2001 using a variety ofmethods such as long-line, raft and on-bottom culture(Kaiser

et al

. 1998; FAO 2003). In seabed cultivation,small ‘seed’ mussels (up to 25 mm in shell length) aredredged from their site of natural settlement and trans-ferred to a cultivation site with good conditions formussel growth. The mussel cultivator will lay mussels atthe density and tidal height that will realize the greatestfinancial return when they are harvested. At high mus-

sel densities multilayering is likely to occur, which willincrease mussel bed complexity. However, the accumu-lation of large amounts of sediment from biodeposi-tion can produce a chemically reducing environmentthat will affect the density or diversity of the associatedanimals (Tsuchiya & Nishihira 1985). At low musseldensities, patches of mussels interspersed with baresubstratum may be produced as the mussels clumptogether, and patch size affects the species richness andnumber of individuals of associated species (Tsuchiya& Nishihira 1985).

In past studies, the effects of mussel density on inver-tebrate assemblages and other environmental parametershave been related to the mussel density encountered atthe time of sampling (Commito 1987; Dittman 1990).However, the mussel density encountered within a bed atthe time of sampling requires careful consideration inview of the fact that the mussel bed will change dynam-ically due to mussel growth and mortality (predationand density-dependent effects). As a result, the infaunalassemblages encountered at the time of sampling mayreflect not only the mussel density at that time but alsothe initial mussel stocking density. The latter may havea long-term influence through the biodeposition thathas occurred prior to invertebrate sampling.

The need to determine the effects of various activitieswithin the coastal zone has arisen due to growing en-vironmental awareness and legislation. It is of particularimportance to the mussel industry because the areasused for seabed cultivation, such as intertidal mudflatsand sandflats, are specifically covered under EuropeanHabitat Conservation Regulations (Council Directive92/43/EEC Annex I). Areas designated as Special Areasof Conservation (SAC) must undergo an appropriateassessment. Therefore, with the growth of the UK musselindustry it is important to establish the impact ofexpanding the areas of subtidal and intertidal mudflaton which mussels are laid.

The present study was carried out at a site of com-mercial mussel cultivation in the Menai Strait, northWales, UK, which is a proposed SAC. The study formedpart of an extensive mussel growth experiment thataimed to determine: (i) any differences in the infaunalcommunity structure between areas of bare mud andareas on which mussels were grown; (ii) any impact ofthe experimental mussel bed on the infaunal commu-nity structure of areas of bare mud both within and sur-rounding the mussel bed, to determine the spatial extentof the potential impact; (iii) the relationship betweenspecies richness and mussel stocking density; (iv) whetherthe infaunal community encountered on the mussel bedat the time of sampling is related to the original densityat which mussels were laid or more closely to the musselpresence at the time of final sampling.

Methods

The study was located on an intertidal mudflat adja-cent to Bangor Pier on the Menai Strait, north Wales

466

H. A. Beadman

et al.

© 2004 British Ecological Society,

Journal of Applied Ecology

,

41

,464–475

(53

°

14

′

N, 04

°

07

′

W), at approximately low water springtide level. Mussels have been cultivated in this area, bylaying directly onto the substratum, since the 1960s(Dare 1980); however, the actual experimental site hadnot been used for this purpose before April 2000 (K.Mould, personal communication).

The experiment was conducted as a part of a large-scalemussel

Mytilus edulis

(L.) growth experiment establishedin October 1999 consisting of a 4

×

4 Latin square of 16individual plots (20

×

20 m). Initially, in October 1999,the infaunal community was sampled by taking fivecylindrical cores (15 cm diameter

×

15 cm deep) at ran-dom from four of the plots within (A3, B4, C2, D5), andfrom the four plots outside, the Latin square (Fig. 1).The samples taken within the Latin square were termedthe 1999 ‘mussel-free area’, and the samples taken outsidewere termed the 1999 ‘distant’ controls. The distant con-trols were used to control for the wider-scale ecologicalvariation that might occur outwith the experimentalarea, from 20 to 80 m distant from the experimentalsite. A sediment sample was taken in each plot using a

cylindrical core (5 cm diameter

×

5 cm deep) and storedfrozen until analysed.

In April 2000 each of the plots in the Latin squarewas seeded with mussels at one of four different stock-ing treatments (7·5, 5, 3 and 2 kg m

−

2

), which meantthat approximately 27 tonnes of mussels were laid intotal. The Latin square was marked with buoys and themussels scattered over each plot from a boat. It wasunfeasible to lay mussels by walking the plots as thiswould have caused more disturbance than the musselcultivation due to the soft nature of the sediment. Dueto the effects of tidal currents and boat positioning, itwas not possible to lay the mussels in precise squares;however, an a posteriori examination walking aroundthe experimental site at low tide revealed that a relativelyeven initial distribution of mussels at the designateddensities had occurred.

In October 2001, 18 months after seeding, the sitewas resampled. Five cores were taken randomly withineach plot from those areas within the experimentalarea that were covered with mussels. In the lowerdensity plots, the mussels had clumped together inlarge patches, thereby leaving areas within each plotthat were not covered with mussels. As a result it was

Fig. 1. Experimental design and location adjacent to Bangor Pier in the Menai Strait, north Wales.

467

Species richness and bivalve stocking density

© 2004 British Ecological Society,

Journal of Applied Ecology

,

41

,464–475

possible to take an extra set of samples from patches ofmussel-free mud from within the lowest mussel treat-ment plots. From direct observation the mussel clumpswere found to form within 2–3 weeks after the musselswere initially seeded, providing areas of bare mud ofrelatively the same size within the low density treatmentplots. The samples taken in these areas of bare mudwere termed the 2001 mussel-free areas, and controlledfor the presence of the mussels within the immediatearea of the experimental plots. Five cores were takenrandomly within the four distant control plots, termedthe 2001 distant controls, and were used as a temporalcontrol. Additionally they provided a control for theeffect of the presence of mussels at a greater spatial scale.A sediment sample was taken in each plot within theexperimental area, after first removing any musselscovering the substratum.

The infaunal samples were washed over a 0·5-mm meshand the residue preserved in 4% formalin. Animals wereidentified to the lowest possible taxonomic level. Thetotal number of individuals and the total number ofspecies for each sample were counted excluding mussels.

Mussels in each sample were counted separately andtheir length was measured, using Vernier callipers, asthe distance between the umbone and the edge of theposterior margin of the shell. Total mussel surface area(length

2

), mussel volume ( length

3

) and dry weight wereestimated for each of the plots. Mussel flesh dry weightwas determined by drying in an oven at 90

°

C for 12 h.The sediment samples were analysed for organic con-tent by drying the sediment at 90

°

C for 12 h and thenincinerating a known weight of dry sediment at 550

°

Cfor 6 h. The percentage organic content was determinedfrom the loss of weight on ignition (Holme & McIntyre1984).

The data from the five cores collected from each plotwere pooled prior to undertaking further analyses. Thedata were analysed in three ways, as described below.

Data were grouped initially according to the pres-ence or absence of mussels, creating three groups: (i)with mussels; (ii) mussel-free areas (2001); (iii) distantcontrols (2001). The

ecological statistical soft-ware package (Clarke & Warwick 1994) was used formultivariate analyses. Cluster analyses on the com-munity data were performed using the Bray–Curtisindex of similarity on fourth root-transformed data,followed by multidimensional scaling (MDS). The overallpercentage contribution of each species to the averagedissimilarity between two groups was ascertained usingSIMPER. Differences between treatments were testedusing an a priori one-way analysis of similarities (

)test. Ranked species abundance plots (dominanceplots) were constructed to analyse changes in communitystructure.

Mann–Whitney pairwise tests were used to test fordifferences in the median number of individuals, median

number of species, and numbers of individuals perspecies between treatments. Differences in the medianorganic content of the sediment samples at each treat-ment were also tested using Mann–Whitney pairwisetests. Non-parametric tests were used, as the data werenot normally distributed (Anderson–Darling test). Datapresented in either tables or figures show means ratherthan medians to aid visual interpretation.

The 1999 mussel-free area and 1999 distant controlplots were compared to determine if there was spatialhomogeneity across the site prior to the seeding of mussels.Comparisons were made of the 1999 and 2001 distantcontrol plots and 1999 and 2001 mussel-free area plotsto ascertain the temporal variation between 1999 and2001. The 2001 mussel-free area and distant controlswere compared to test if the potential impact of theexperimental mussel bed on the infaunal communityoperated at different spatial scales. The data were sub-jected to the multivariate community analyses outlinedabove. The abundance of individual taxa and sedimentcharacteristics were treated as above.

In order to analyse specifically the effects of musseldensity upon the associated infaunal community withinthe mussel bed, the control data were excluded from theanalysis. The data were grouped according to the fouroriginal seeding treatments. Cluster analyses againinvolved the use of the Bray–Curtis index of similarityon fourth root-transformed data followed by MDS.Significant differences between treatments were deter-mined using an a priori

test. Kruskal–Wallistests were used to test for significant differences in themedian number of individuals, the median number ofspecies between treatments, and differences in the medianorganic content of the sediment samples at each treatment.

The mussel density at the time of sampling was notgrouped

post hoc

. The data were analysed according tothe actual number of mussels in each plot. Counts ofmussels were excluded from the multivariate com-munity analyses and hence were treated as a variableagainst which the community relationships could becompared. The relationship between the environ-mental factors, in terms of mussel presence (number, area,volume, dry weight) and sediment organic content, andthe benthic community were investigated using

and the

test (Clarke & Warwick 1994).

compares the similarity matrix generated using com-munity data with a similarity matrix generated usingenvironmental parameters for the same sites or repli-cate treatments. The mussel parameters were con-sidered to be ‘environmental’ parameters as mussels werenot included in the community data set.

is usedto match biotic to environmental patterns in data. Theprocedure considers increasing levels of complexity,i.e. single variables then combinations of two variables,etc. The variable or combination of variables that bestexplains the patterns observed in the biological data setis identified. The significance of the correlation betweenthe biological and environmental data was tested usingthe

permutation test. This tests the hypothesis

468

H. A. Beadman

et al.

© 2004 British Ecological Society,

Journal of Applied Ecology

,

41

,464–475

that there is no relationship between the biologicalinformation and a specific abiotic/environmental pattern,i.e. that

ρ

w

is zero. This is examined by a randomizationtest in which

ρ

w

is recomputed for all permutations ofthe samples in one of the two underlying similaritymatrices. If the observed value of

ρ

w

exceeds that foundin 95% of the simulations, then the null hypothesiscan be rejected at the 5% level. Regression analysiswas used to determine whether significant relationshipsoccurred between mussel area and total number ofspecies, total abundance of individuals (all taxa pooled)and the abundance of individual species. The data wastested for normality using the Anderson–Darling test.

Results

There was a significant difference between the infaunalcommunities of both of the treatments without mussels(the mussel-free area 2001 and distant control 2001) andthe areas of mudflat on which mussels were present.(

;

R

= 0·830,

P

< 0·001;

R

= 0·934,

P

< 0·001,respectively; Fig. 2). These differences were partlyexplained for the distant controls by a greater numberof animals per replicate and a higher number of speciesper replicate compared with the plots with mussels(Mann–Whitney;

W

= 144·0, d.f. = 14,

P

< 0·05;

W

=141·0, d.f. = 14,

P

< 0·05, respectively; Fig. 3). Themedian abundance and number of species in the plots

with mussels were not significantly different from the2001 mussel-free area (Mann–Whitney;

W

= 153·0,d.f. = 14,

P

= 0·171;

W

= 152·0, d.f. = 14,

P

= 0·157,respectively) (Fig. 4). However, the identity of some ofthe species recorded was different between the twotreatments (Table 1). There was no clear difference incommunity structure between the treatments in termsof dominance by rank species abundance.

An analysis of the individual species that contributedto the major difference between the 2001 treatmentswith no mussels (mussel-free area and distant control)and plots with mussels illustrated the differences in theinfaunal communities. For example,

Pseudopolydoraantennata

and

Pygospio elegans

(only one individualwas present in the lowest density mussel sample) wereonly present in the control plots. Conversely,

Carcinusmaenas

and

Scolelepis squamata

were only foundwithin the plots with mussels (mean

±

SE, 8·56

±

1·22and 0·94

±

0·87, respectively). Other taxa showedeither increased (Cirratulidae,

Corophium

spp.,

Nephtyshombergii

,

Notomastus latericeus

) or decreased (

Melitapalmata

,

Tubificoides benedeni

) numbers of individualsin both 2001 control plots compared with the plotswith mussels (Fig. 5 and Table 2).

Organic content of the sediment was significantlyhigher in the plots with mussels than in either of the2001 treatments with no mussels (Mann–Whitney;

W

=200 d.f. = 14

P

< 0·05 in both cases).

Fig. 2. Two-dimensional MDS ordination of community data found in the controls, mussel-free areas and plots with mussels2001 (stress = 0·1).

Fig. 3. Mean abundance of infaunal animals in the 2001 distantcontrols, mussel-free areas and plots with mussels (± SE).Means and SE are shown for clarity rather than median values.

Fig. 4. Mean number of species of infaunal animals in the 2001distant controls, mussel-free areas and plots with mussels(± SE). Means and SE are shown for clarity rather thanmedian values.

469

Species richness and bivalve stocking density

© 2004 British Ecological Society,

Journal of Applied Ecology

,

41

,464–475

There was no significant difference between theinfaunal communities of the treatments with nomussels in 1999 prior to seeding with mussels (

;

R

= 0·031,

P

= 0·42; Fig. 2). In addition, there was nosignificant difference in the total number of individualsand number of species in the mussel-free area anddistant control areas in 1999 (Mann–Whitney;

W

= 14·0,d.f. = 6,

P

= 0·312;

W

= 21·0, d.f. = 6,

P

= 0·471, respect-ively). There was no significant difference in the organic

content of the treatments with no mussels in 1999(Mann–Whitney;

W

= 19·0, d.f. = 6,

P

= 0·885).The infaunal communities of the mussel-free areas

taken in 1999 and 2001 were significantly different(

;

R

= 0·854,

P

< 0·05; Fig. 2). This was reflectedin both the numbers of individuals and number of species,which were significantly lower in the 2001 samples(Mann–Whitney;

W

= 26·0, d.f. = 5,

P

< 0·05 in bothcases). Of a total of 29 species found in the plots in

Table 1. Taxa recorded within the mussels plots, in the 2001 mussel-free area, and 2001 distant controls (sampling effort,respectively, n = 16, n = 4, n = 4)

Taxa Class Mussel plots 2001 mussel-free area 2001 distant control

Abra spp. Lamarck, 1818 Pe +Anemone spp. A +Amphicteis gunneri (M Sars, 1835) P +Amphipholis squamata (Chiaje, 1829) Op +Carcinus maenas (Linnaeus, 1758) M +Gammarus locusta (Linnaeus, 1758) M +Macoma balthica (Linnaeus, 1758) Pe +Malacoceros fuliginosus (Claparède, 1868) P +Modiolula phaseolina (Philippi, 1844) Pe +Mysella bidentata (Montagu, 1803) Pe +Mytilus edulis (Linnaeus, 1758) Pe +Nemertea N* +Oligochaeta O +Pholoe assimilis Oersted, 1845 P +Pinnotheres pisum (Linnaeus, 1767) M +Scolelepis squamata (Abildgaard, 1806) P +Sthenelais boa (Johnston, 1839) P +Pseudomystides limbata (Saint-Joseph, 1888) P + +Melita palmata (Montagu, 1804) M + + +Ampharete acutifrons Grube, 1860 P + + +Capitellides spp. Mesnil, 1897 P + + +Capitomastus spp. Eisig, 1887 P + + +Cirratulidae spp. P + + +Corophium spp. Latreille, 1806 M + + +Mysta picta (Quatrefages, 1866) P + + +Nephtys hombergii Savigny, 1818 P + + +Hediste diversicolor (O F Müller, 1776) P + + +Notomastus latericeus M Sars, 1851 P + + +Anaitides maculata (Linnaeus, 1767) P + + +Pholoe inornata Johnston, 1839 P + + +Scoloplos armiger (O F Müller, 1776) P + + +Tubificoides benedenii (Udekem, 1855) O + + +Glycera spp. Savigny, 1818 P + +Galathowenia oculata Zaks, 1922 P + +Pygospio elegans Claparède, 1863 P + +Maldanidae P +Nephtys ( juvenile) spp. Cuvier, 1817 P +Nereimyra punctata (O F Müller, 1788) P + +Ampharetidae P + +Gastrosaccus sanctus (van Beneden, 1861) M +Lanice conchilega (Pallas, 1766) P +Nephtys assimilis Oersted, 1843 P +Nephtys kersivalensis McIntosh, 1908 P +Nucella spp. Röding, 1798 G +Owenia fusiformis Chiaje, 1842 P +Pseudopolydora antennata (Claparède, 1870) P +Sphaerodoropis balticum (Reimers, 1933) P +Spiophanes bombyx (Claparède, 1870) P +Sthenelais limicola (Ehlers, 1864) P +

Class: A, Anthozoa; G, Gastropoda; M, Malacostraca; O, Oligochaeta; Op, Ophiuroidea; P, Polychaeta; Pe, Pelecypoda.*N, phylum Nemertea.

470H. A. Beadman et al.

© 2004 British Ecological Society, Journal of Applied Ecology, 41,464–475

1999, 17 species were not found in these plots in 2001and these species contributed to more than 40%() of the dissimilarity between the mussel-freearea plots in 1999 and 2001. A further 20% of the dis-similarity could be attributed to those species thatoccurred in both 1999 and 2001 (Fig. 5). However,of these taxa only Corophium spp. had significant

differences in numbers (Mann–Whitney; W = 10·0, d.f.= 6, P < 0·05) and Melita palmata was only present in2001. There was no significant difference in the organiccontent of the mussel-free area plots in 1999 and 2001(Mann–Whitney; W = 24·0, P = 0·112).

The distant controls did not change significantlybetween 1999 and 2001 in either the infaunal community

Fig. 5. Mean number of individuals (± SE) of taxa contributing to the major differences between infaunal communities of the2001 mussel-free areas (MFA), distant controls (DC) and plots with mussels. Means and SE are shown for clarity rather thanmedian values.

Table 2. Mann–Whitney tests (d.f. = 18) of the individual taxa that contributed to the major community difference betweeneither 2001 distant controls or 2001 mussel-free areas and plots with mussels

Taxa Plots with mussels compared with W P

Cirratulidae Distant control 2001 19·0 0·885Mussel-free area 2001 151·0 0·119

Corophium spp. Distant control 2001 136·0 0·003Mussel-free area 2001 141·0 0·012

Nephtys hombergii Distant control 2001 138·5 0·061Mussel-free area 2001 159·0 0·421

Notomastus latericeus Distant control 2001 137·0 0·004Mussel-free area 2001 146·5 0·047

Melita palmata Distant control 2001 192·5 0·021Mussel-free area 2001 191·0 0·033

Tubificoides benedenii Distant control 2001 178·5 0·345Mussel-free area 2001 194·0 0·003

471Species richness and bivalve stocking density

© 2004 British Ecological Society, Journal of Applied Ecology, 41,464–475

(; R = 0·448, P = 0·06) (Fig. 2), the total numberof individuals or the number of species (Mann–Whitney; W = 22·0, d.f. = 6, P = 0·312; W = 17·5, d.f. = 6,P = 1·00, respectively). There was no significant dif-ference in the organic content of the distant controltreatments in 1999 and 2001 (Mann–Whitney; W =18·0, P = 0·885).

The infaunal communities of the mussel-free areaand distant control treatments sampled in 2001 weresignificantly different (; R = 0·635, P = 0·03) butthere was no significant difference in the total numberof individuals between these treatments (Mann–Whitney;W = 13·0, d.f. = 6, P = 0·194). However, there was asignificantly lower number of species found per repli-cate plot for the mussel-free area vs. the distant controltreatment (Mann–Whitney; W = 10·0, d.f. = 6, P < 0·05).There was no significant difference in the organic con-tent of the sediment in the mussel-free area and distantcontrol in 2001 (Mann–Whitney; W = 24·0, P = 0·112).

The infaunal communities associated with plotswithin the Latin square, when grouped according tothe four original seeding densities laid in April 2000,were not significantly different from each other (Fig. 2;; R = 0·133, P = 0·061). Similarly, there were nosignificant differences in the number of individuals orthe number of species (Kruskal–Wallis; H = 1·35, d.f. =11, P = 0·718; H = 6·77, d.f. = 12, P = 0·08, respec-tively), and there was no significant difference in theorganic content of the sediment for the four differentmussel density treatments (Kruskal–Wallis; H = 0·11,d.f. = 11, P = 0·991).

indicated that there was a highly significantcorrelation between the environmental and biologicaldata (ρ = 0·513, P = 0·003). demonstrated thatthe strongest correlation between infaunal communityand environmental data was with mussel shell areaand mussel volume (; ρ = 0·643; ρ = 0·642,respectively).

The gradient between the infaunal community andmussel shell area was supported by the relationshipsbetween both abundance of individuals and number ofspecies present in the samples plotted against musselarea (Table 3). The number of individuals per plot[ln(n + 1)] had a significant negative linear relationshipwith mussel area (Fig. 6). The number of species alsohad a significant negative linear relationship withln(n + 1) mussel area (Fig. 7). Analysis of the mean

abundance [ln(n + 1)] of individual species revealedsignificant negative relationships with mussel area forthree taxa: Cirratulidae, Corophium spp. and Melitapalmata (Table 3 and Fig. 8).

Discussion

As in other studies (Commito 1987; Dittman 1990;Guenther 1996; Ragnarsson & Raffaelli 1999), it isclear that mussel beds affect the composition, richnessand abundance of the benthic fauna of the sedimentonto which they are laid, in terms of both the numbersof individuals and species present (Figs 3 and 4 andTable 1). In this large-scale experiment the presenceof mussels had a large impact on the abundance ofepibenthic crustaceans, in particular Carcinus maenas

Table 3. Regression analysis of infaunal community data with area of mussels per sample [except number of species, which isregressed against ln(n + 1)mussel area]. Numbers of individuals are ln(n + 1)

Mussel area (V)Correlation coefficient (r) d.f. P Slope (± SE) Intercept (± SE)

Coefficient of determination (r2)

Number of individuals per sample 0·64 13 0·005 −0·000365 ± 0·000106 6·77 ± 0·370 0·404Number of species per sample 0·65 13 0·004 −5·48 ± 1·56 53·3 ± 12·5 0·425Cirratulidae per sample 0·74 13 0·001 −0·000858 ± 0·000196 7·22 ± 0·684 0·548Corophium spp. per sample 0·60 13 0·009 −0·000473 ± 0·000155 1·99 ± 0·541 0·357Melita palmata per sample 0·67 13 0·002 −0·000547 ± 0·000148 3·59 ± 0·515 0·459

Fig. 6. Relationship between area of mussels and number ofindividuals. Values of 2001 distant controls and mussel-freeareas are shown for comparison. Regression line y = 6·77 –0·000365x, r 2 = 0·404.

Fig. 7. Relationship between area of mussels and number ofspecies in sample. Values of 2001 distant controls and mussel-free areas are shown for comparison. Regression line y = 53·3 –5·48x, r 2 = 0·425.

472H. A. Beadman et al.

© 2004 British Ecological Society, Journal of Applied Ecology, 41,464–475

and Melita palmata. This can be attributed to the refugeprovided by the mussel matrix from water movement,desiccation and predation (Dittman 1990; Ragnarsson& Raffaelli 1999). Commito & Boncavage (1989) sug-gested that the presence of mussels also caused anincrease in oligochaete abundance. This concurs withour findings in which the abundance of Tubificoidesbenedeni was greater in the mussel plots compared withareas with no mussels (Fig. 5). The presence of musselson soft sediments has been associated with a shift in thecommunity from one dominated by polychaetes to onedominated by oligochaetes (Commito 1987; Commito& Boncavage 1989; Dittman 1990). Although such atrend was not as apparent here, there was a suggestionthat samples with more mussels were associated with areduced abundance of cirratulids (Fig. 8) while therewas no concomitant change in the abundance of oligo-chaetes. Thus the latter became more dominant interms of their overall contribution to faunal composition.The high abundance of Tubificoides benedeni in musselbeds has been attributed to its tolerance of organicallyrich deoxygenated sediment (Commito & Boncavage1989). Its reproductive strategy also overcomes the

problem of ingestion by mussel filtration due to theproduction of non-larval benthic offspring from cocoons(Hunter & Arthur 1978).

While some species increased in numbers in thepresence of mussels, other species decreased. Pygospioelegans was less abundant in the mussel bed infaunalcommunity than in surrounding sediment and this hasalso been demonstrated in the Wadden Sea (the Neth-erlands) and the Ythan Estuary (Scotland) (Guenther1996; Ragnarsson & Raffaelli 1999). A decline in thenumber of Pygospio elegans has been attributed tounstable sediments (Wilson 1981; Flach 1996), whicharise in a mussel bed due to the high deposition rates offaeces and pseudofaeces, and the movement of themussels themselves, which may cause tube destruction(Kautsky & Evans 1987). Corophium spp., a burrow-dwelling invertebrate, also declined in numbers in themussel bed and this again can be attributed to theunstable sediment regime (Jensen & Kirstensen 1990).Mussel beds may prove a less suitable habitat for suchorganisms simply due to lack of space for burrowconstruction as well as the movements and growthof adjacent mussels that impinge upon burrows. The

Fig. 8. Relationship between surface area of mussels and the associated number of selected taxa.

473Species richness and bivalve stocking density

© 2004 British Ecological Society, Journal of Applied Ecology, 41,464–475

other species that showed a decline in numbers in themussel bed (Fig. 5) reflected both the physical environ-ments of the mussel bed and the associated infaunalcommunity. The capitellid Notomastus latericeus pre-fers cleaner muddy sand and is a more selective feederthan the other, more opportunistic, capitellid species,and this was reflected by its higher abundance in bothof the control treatments compared with the musseltreatment (Fauchald & Jumars 1979). The decliningnumbers of the carnivorous predator Nephtys hombergiimay reflect the reduced total abundance of individuals(and hence prey) in the mussel bed compared with thesurrounding sediments.

The distant control plots represented an area closeto the mussel beds (20–80 m distant) compared withthe mussel-free area plots that were among the musselbed treatments. Examination of this spatial gradientsuggests that the effects of mussels on benthic infaunalcommunities of soft sediments are reduced withincreasing distance from the mussel bed, and there wasno significant effect on the infaunal community of thedistant controls. Although natural changes over timemay account for some of the differences between the1999 and 2001 treatment plots with no mussels, themagnitude of the difference in species compositionsuggests that this was not the main factor. This wassupported by the lack of significant change in thecommunity of the distant controls over time. Prior tothe seeding of the mussel bed the spatial homogeneity ofthe site was confirmed by comparison of the 1999 plotswith no mussels, hence the distant controls and mussel-free area plots were directly comparable. It is likely thatthe clear patterns with increasing distance from themussel bed treatments indicated a steep dilution of theinfluence of the mussel bed on the benthic community.Thus it would appear that the influence of the musselbed on benthic invertebrate communities is undetectableat a distance of between 20 m and 80 m from the musselbed. This effect is demonstrated in both the medianabundance and the median species number in eachtreatment. Dittman (1990) demonstrated a reducedabundance of individuals within a mussel bed com-pared with the surrounding sediment, which concurswith our study, with a reduced abundance number ofindividuals with increasing proximity to the mussel bed(Fig. 3) (cf. Commito 1987). However, the number ofspecies was not significantly different between themussel bed and surrounding area in Dittman’s study(1990). This does not concur with our results, in whichthe numbers of species found in the mussel bed werelower than in the 2001 distant control treatment plots.Although it appears from Table 1 that more speciesoccurred overall in the mussels plots, the sampling effortis four times greater that for the plots with no mussels.

The treatments with no mussels (mussel-free area anddistant control) did not differ in terms of their physicalenvironment with respect to the organic content of thesediment, hence sediment conditions do not appear tobe the cause of the observed community differences.

Multivariate analysis revealed that the mussel-free areaand distant control areas were significantly different in2001, largely attributed to a significantly lower mediannumber of species in the mussel-free area comparedwith the distant control. It may be that the larvae ofcertain species of the infaunal community found inthe vicinity of the mussel bed are more susceptible toremoval through bivalve filtration, either through sizeselection by the mussels or due to some behaviouralcharacteristics of different invertebrate larvae. Woodin(1976) suggested that suspension-feeding bivalvescould have a negative effect on the recruitment of infau-nal species due to predation by filter feeding, althoughthis hypothesis was refined by Commito & Boncavage(1989) to preclude organisms that do not have a pelagicdevelopment stage (e.g. Tubificoides benedeni). A studyconducted at a much smaller spatial scale (1-m2 experi-mental plots) did not detect a significant effect of bivalvedensity on larval settlement and juvenile recruitment(Hunt, Ambrose & Peterson 1987). Nonetheless, at thelarge scale used in this study filtration by the mussel bedis likely to have an effect on the benthic infaunal com-munity within the bed, either through larval removal orthrough competition for food (Cowden, Young & Chia1984; Morgan 1992).

The effects of a mussel bed on its associated benthiccommunity were variable within the bed, as demon-strated by significant relationships between the benthiccommunity parameters and the area covered bymussels in individual replicates. These relationshipsreflected the area of mussels at the time of samplingrather than the history of the mussel bed in terms of theinitial stocking treatment. This suggests that the com-position of the associated benthic community is closelylinked to mussel density as it changes through time. Norelationship could be found between the benthiccommunities and the original seeding treatment, andno lasting effect on the organic content of the sedimentdue to mussel treatment was detected. At the time ofsampling, the lowest area of mussel cover was associ-ated with the highest number of species compared withthe 2001 controls and higher mussel densities (Fig. 7).These areas of low mussel cover are capable of support-ing a greater number of species, as habitats suitablefor both the mudflat fauna and mussel bed fauna areprovided by the extra microhabitats provided withinisolated clumps of mussels. However, as the area of sub-stratum covered by mussels increased, a negative rela-tionship occurred for both the abundance of individualsand the number of species (Figs 6 and 7). This suggeststhat the negative factors of a highly anoxic environment,competition for food and space, and the filtration ofpelagic larvae that occurred in areas of high musselcoverage, outweigh the more positive benefits ofincreased habitat complexity and refugia providedwithin the mussel bed matrix. Similar responses ofinvertebrate species to increasing bivalve density havebeen reported elsewhere. For example, Spencer, Kaiser& Edwards (1996) reported a linear decrease in the

474H. A. Beadman et al.

© 2004 British Ecological Society, Journal of Applied Ecology, 41,464–475

number of cirratulids with increasing bivalve density inplots of cultivated Manila clams (Tapes philippinarumAdams & Reeve 1950).

Mussel beds can alter the infaunal benthic communityof the adjacent and interstitial sediments. This studyhas demonstrated that this results not only in a changein the composition of species of the infaunal commu-nity, but also the number of individuals and number ofspecies. At all but the lowest areas of mussel cover, theinfaunal communities of plots with mussels had fewerindividuals and fewer species. Within the mussel beditself negative trends of species numbers and abund-ance of individuals with increased mussel shell areawere also demonstrated. Furthermore, the data sug-gested that the effects of mussel beds on the infaunalcommunities of surrounding sediments were localizedto 0–10 m, and no significant effect occurred at a scaleof 20–80 m. However, although the species composi-tion and abundance of individual invertebrate speciesmay be altered by the presence of mussels, the distribu-tion of individuals among species remained relativelyunchanged.

Expanding the extent of present seabed musselcultivation will affect the invertebrate assemblage ofthe sediments on which the mussels are laid. However,changes in benthic community composition could bereduced by proportional lowering of the stockingdensity of mussels within mussel beds or by restrictionof the final surface area of coverage of the seabed. Itwould also appear that in intertidal areas, the effectsof seabed cultivation of mussels are restricted to theimmediate area of cultivation and that wider-scaleeffects are not apparent. Therefore, controlling the areaavailable for cultivation would be an effective manage-ment measure to reduce impacts on benthic fauna.

Acknowledgements

This study was funded through a Natural Environ-mental Research Council LINK-Aquaculture awardENV10. The authors are grateful to A. Wilson of DeepDock Ltd and K. Mould of Myti Mussels Ltd fortheir assistance in the execution of this project. We alsothank A. Navarro and K. Dernie for their help in thefield and laboratory.

References

Chamberlain, D.E., Fuller, R.J., Bunce, R.G.H., Duckworth, J.C.& Shrubb, M. (2000) Changes in the abundance of farmlandbirds in relation to the timing of agricultural intensificationin England and Wales. Journal of Applied Ecology, 37, 771–788.

Clarke, K.R. & Warwick, R.M. (1994) Change in MarineCommunities: An Approach to Statistical Analysis and Inter-pretation. National Environmental Research Council,Swindon, UK.

Commito, J.A. (1987) Adult–larval interactions: predictions,mussels and cocoons. Estuarine Coastal and Shelf Science,25, 599–606.

Commito, J.A. & Boncavage, E.M. (1989) Suspension-feeders and co-existing infauna: an enhancement example.Journal of Experimental Marine Biology and Ecology, 125,33–42.

Costanza, R., D’Arge, R., De Groot, R., Farber, S., Grasso, M.,Hannon, B., Limburg, K., Naeem, S., O’Neill, R.V.,Paruelo, J., Raskin, R.G., Sutton, P. & Van den Belt, M.(1997) The value of the world’s ecosystem services andnatural capital. Nature, 387, 253–260.

Cowden, C., Young, C.M. & Chia, F.S. (1984) Differentialpredation on marine invertebrate larvae by two benthicpredators. Marine Ecology Progress Series, 14, 145–149.

Dare, P.J. (1980) Mussel Cultivation in England and Wales.Laboratory Leaflet. Ministry of Agriculture, Fisheries & Food,Directorate of Fisheries Research, Lowestoft, UK.

Dittman, S. (1990) Mussel beds – amensalism or ameliorationfor intertidal fauna? Helgoland Marine Research, 44, 335–352.

FAO (2003) FAOSTAT database (last updated 29 Aug 2003).Food and Agriculture Organisation of the United Nations,Rome (http://www.fao.org)

Fauchald, K. & Jumars, P.A. (1979) The diet of worms: astudy of Polychaete feeding guilds. Oceanography andMarine Biology: An Annual Review, 17, 193–284.

Flach, E.C. (1996) The influence of the cockle, Cerastodermaedule, on the macrozoobenthic community of tidal flats inthe Wadden Sea. Marine Ecology, 17, 87–98.

Guenther, C.P. (1996) Development of small Mytilus bedsand its effects on resident intertidal macrofauna. MarineEcology, 17, 117–130.

Herman, P.M.J. (1993) A set of models to investigate therole of benthic suspension feeders in estuarine ecosystems.Bivalve Filter Feeders in Coastal Ecosystem Processes(ed. R.F. Dame), pp. 421–454. Springer-Verlag, Berlin,Germany.

Holme, N.A. & McIntyre, A.D. (1984) Measurement of thePhysical and Chemical Environment. Methods for the Studyof Marine Benthos. Blackwell Scientific Publications,Oxford, UK.

Hunt, J.H., Ambrose, W.G.J. & Peterson, C.H. (1987) Effectsof the gastropod, Ilyanassa obsoleta (Say), and the bivalve,Mercenaria mercenaria (L.), on larval settlement andjuvenile recruitment of infauna. Journal of ExperimentalMarine Biology and Ecology, 108, 229–240.

Hunter, J. & Arthur, D.R. (1978) Some aspects of the ecologyof Peloscolex benedeni Udekem (Oligochaeta: Tubificidae)in the Thames Estuary. Estuarine, Coastal and Shelf Science,6, 197–208.

Jensen, K.T. & Kirstensen, L.D. (1990) A field experiment oncompetition between Corophium volutator (Pallas) andCorophium arenarium (Crawford) (Crustacea: Amphipodia):effects on survival, reproduction and recruitment.Journal of Experimental Marine Biology and Ecology, 137,1–24.

Kaiser, M.J., Laing, I., Utting, S.D. & Burnell, G.M. (1998)Environmental impacts of bivalve mariculture. Journal ofShellfish Research, 17, 59–66.

Kautsky, N. & Evans, S. (1987) Role of biodeposition byMytilus edulis in the circulation of matter and nutrients in aBaltic coastal system. Marine Ecology Progress Series, 38,201–212.

Marshall, E.J.P., Brown, V.K., Boatman, N.D., Lutman, P.J.W.,Squire, G.R. & Ward, L.K. (2003) The role of weeds insupporting biological diversity within crop fields. WeedResearch, 43, 77–89.

Morgan, S.G. (1992) Predation by planktonic and benthicinvertebrates on larvae of estuarine crabs. Journal of Experi-mental Marine Biology and Ecology, 163, 91–110.

475Species richness and bivalve stocking density

© 2004 British Ecological Society, Journal of Applied Ecology, 41,464–475

Naylor, R.L., Goldburg, R., Primavera, J., Kautsky, N.,Beveridge, M.C.M., Clay, J., Folke, C., Lubchenco, J.,Mooney, H.A. & Troell, M. (2000) Effect of aquaculture onworld fish supply. Nature, 405, 1017–1024.

Ragnarsson, S.A. & Raffaelli, D. (1999) Effect of the musselMytilus edulis on the invertebrate fauna of sediments.Journal of Experimental Marine Biology and Ecology, 241,31–43.

Seed, R. (1976) Ecology of marine mussels. Marine Mussels:Their Ecology and Physiology (ed. B.L. Bayne), pp. 13–65.Cambridge University Press, London.

Seed, R. & Suchanek, T.H. (1992) Population and communityecology. The Mussel Mytilus: Ecology, Physiology, Geneticsand Culture (ed. E. Gosling), pp. 87–157. Elsevier, Amster-dam, the Netherlands.

Spencer, B.E., Kaiser, M.J. & Edwards, D.B. (1996) Theeffect of Manila clam cultivation on an intertidal benthiccommunity: the early cultivational phase. Aquaculture Re-search, 27, 261–276.

Symstad, A.J., Siemann, E. & Haarstad, J. (2000) An experi-mental test of the effect of plant functional group diversityon arthropod diversity. Oikos, 89, 243–253.

Tsuchiya, M. (1980) Biodeposition production by the mussel

Mytilus edulis L. on rocky shores. Journal of ExperimentalMarine Biology and Ecology, 47, 203–222.

Tsuchiya, M. & Nishihira, M. (1985) Islands of Mytilus ashabitat for small intertidal animals: effect of island size oncommunity structure. Marine Ecology Progress Series, 25,71–81.

Tsuchiya, M. & Nishihira, M. (1986) Islands of Mytilus ashabitat for small intertidal animals; effects of Mytilusage structure on the species composition of the associatedfauna and community organisation. Marine Ecology ProgressSeries, 31, 171–178.

Wahl, M. (2001) Small scale variability of benthic assemblages:biogenic neighborhood effects. Journal of ExperimentalMarine Biology and Ecology, 258, 101–114.

Wilson,W.H. Jr (1981) Sediment-mediated interactions ina densely populated infaunal assemblage: the effects ofthe polychaete Abarenicola pacifica. Journal of MarineResearch, 39, 735–748.

Woodin, S.A. (1976) Adult–larval interactions in denseinfaunal assemblages: patterns of abundance. Journal ofMarine Research, 34, 25–41.

Received 23 May 2003; final copy received 29 January 2004